Download
1 variant available
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.
Frieren LoRA - Sousou no Frieren Universe Project
📌 If you’re interested in my larger meta-universe project (multiple styles, full universes, 20+ personalized characters), please read this post to the end.
Recommended Prompt Structure
anime, friStyle, a full body portrait of young elf woman friFrieren. She has a petite figure with small breasts, pale skin, turquoise eyes, long pointed elf ears, and lavender-white pigtails with red drop earrings. friFrieren wears a white mage tunic with gold trim, a striped collar shirt, dark tights, and brown boots. friFrieren is standing with a neutral expression. White plain background.
Trigger words / core tags:
- anime: base 2D anime style
- friStyle: universe-specific style tag (will become very useful when combining multiple universe LoRAs)
- friFrieren: character trigger (still recommended to add basic appearance description)
- anatomy: petite figure with small breasts, pale skin, turquoise eyes, long pointed elf ears, and lavender-white pigtails with red drop earrings
- outfit: white mage tunic with gold trim, a striped collar shirt, dark tights, and brown boots
About this LoRA
This is a strong, dedicated character LoRA trained to capture Frieren from Sousou no Frieren with high accuracy. It faithfully reproduces her canonical look:
The dataset contains both anime/friStyle and photorealistic images. Thanks to Flux 2 Klein’s capabilities, you can comfortably experiment with other artistic styles even without additional LoRAs.
The gallery includes an example of compatibility with “gnStyle lora”, a Lora set in the Gnossia universe. This Lora isn't currently available in the release and likely won't be (unless it's in the form of a style, without specific characters) until the larger Lora I mention below is released.
NSFW note: The training dataset includes nude photos of Freiren, so it works well with this LoRA to generate advanced scenes. I highly recommend it if you're looking for high-quality NSFW results.
Current Project & Future Plans (Sousou no Frieren Universe)
This LoRA is the first release in a dedicated Sousou no Frieren universe pack.
Planned additions:
- Ubel
- Young Himmel
- Stark
- Fern
- Backgrounds / locations
- Secondary characters (without personalization)
Bigger Picture – Meta-Universe Project
This Frieren universe is only one small part of a much larger meta-project. It will eventually include:
- Reworked versions of kyoanistyle and a1 style
- 4-5 additional full universes
- ~20 carefully selected personalized characters
The meta-project already has over 1000 personalized images in the training pipeline, but it will take significant time to complete. That’s why releases are currently slower and focused on 1–2 characters per LoRA.
Why I’m building such a large interconnected system:
- Flux Klein still has noticeable issues with true multi-character scenes.
- Concept transfer (clothing, poses, weapons, backgrounds, anatomy quirks) between styles and characters is extremely powerful.
- With a well-built base of ~100 images, new characters can often be trained successfully with only 8–12 dedicated images (10k steps).
Tip for combining characters:
Because I can’t train datasets larger than 100 images yet, I recommend using Edit + Inpaint sequentially when combining multiple character LoRAs. This gives the cleanest results with minimal or zero bleeding. (I’m aware of FreeFuse but personally don’t like the output quality.)